现在的位置: 首页研究点评, 进展交流>正文
[JAMA发表述评]:将临床试验结果应用于疫情流行期间的临床实践
2023年08月16日 研究点评, 进展交流 [JAMA发表述评]:将临床试验结果应用于疫情流行期间的临床实践已关闭评论

Editorial 

July 10, 2023

Translating Clinical Trial Results to Clinical Practice During a Pandemic

Andre C. Kalil, Angela Huttner, Carlos A. Gomez

JAMA. Published online July 10, 2023. doi:10.1001/jama.2023.11260

In February 2020, the National Institutes of Health (NIH) launched a public-private partnership, the Adaptive COVID-19 Treatment Trial (ACTT) network,1,2 followed in April 2020 by the Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV) network,3 to develop a multidisciplinary coordinated research response to COVID-19. These networks achieved remarkable success in completing multiple adaptive randomized clinical trials (RCTs) and identifying effective (and ineffective) treatments for patients with COVID-19. Even as the pandemic enters a new phase, the results reported from these adaptive trials have important ramifications for current and future investigations of novel therapeutics during a pandemic4 as well as for clinical care, because the risk of serious illness when hospitalization for COVID-19 is required remains elevated among both vaccinated and nonvaccinated individuals.5

In this issue of JAMA, O'Halloran and colleagues3 report the findings of ACTIV-1, a placebo-controlled, masked, randomized clinical trial among patients hospitalized for COVID-19, which compared 3 immunomodulating agents (abatacept, infliximab, or cenicriviroc) with placebo. It included 1971 participants randomized across the 3 substudies in 95 hospitals in the US and Latin America from October 2020 to December 2021.

The primary outcome of time to recovery based on an 8-point ordinal scale was not significantly different for abatacept (recovery rate ratio [RRR], 1.12 [95% CI, 0.98-1.28]; P = .09), cenicriviroc (RRR, 1.01 [95% CI, 0.86-1.18]; P = .94), or infliximab (RRR, 1.12 [95% CI, 0.99-1.28]; P = .08) compared with placebo. Because of a predefined gatekeeping approach, 28-day mortality and day 14 clinical status (key secondary outcomes) could not be concluded to be statistically significant, although these end points are clinically relevant to patients and the differences between placebo were better for both abatacept and infliximab (ie, the 95% CI did not cross the unity).

Notably, the study by O'Halloran et al not only had an appropriate and effective randomization process, but also included a placebo group, which further minimized biases that cannot be prevented by randomization only, such as performance bias, which occurs when the processes of care differ due to knowledge of intervention assignment, a type of bias that is particularly common in acute diseases.6 Specific to COVID-19 trials, the postrandomization differences in standards care (medications and supportive care), varying access to respiratory care (supplemental oxygen and ventilation devices), and differences within health care sites (hospital and intensive care unit beds) without placebo and double-masking procedures could potentially bias mortality in either direction due to the unmeasured and unpredictable multitude of concurrent interventions and local resources. During a pandemic, these postrandomization biases could have made the final results of unmasked trials unreliable regardless of how large the trial sample size was or what the study outcome was (primary and secondary), including mortality, because limited and heterogeneous access to standard care or life-saving cointervention resources cannot be prevented by the randomization process. These issues remain highly relevant today and for future pandemics. This laudable goal of preventing common postrandomization biases that can have major influence in both morbidity and mortality during a pandemic, such as performance and ascertainment biases, by conducting placebo-controlled RCTs during the COVID-19 pandemic has been successfully achieved by several trial networks both in high-income (eg, ACTT1,2) and low- to middle-income (eg, BRICNet,7 STOP-COVID8) countries.

How should the discrepancy between the primary and secondary outcomes in the study by O'Halloran et al be interpreted? The choice of primary and secondary outcomes is one of the most challenging and important decisions made during the planning of a trial. The primary outcome not only needs to be directly relevant to clinical care, but also needs to be attainable. If either of these 2 factors is not met, the trial’s results are unlikely to be implemented, even if the intervention proves to be effective.

The study by O'Halloran et al3 did not detect time-to-recovery (primary outcome) benefits, but detected a reduction in mortality at 28 days (secondary outcome). Here is where the issue gains complexity. The positive secondary outcome results may represent a true benefit that was not detected by the design of the primary outcome or it may represent a false benefit due to multiplicity, a statistical term indicating that the more statistical comparisons are made, the more false-positive results are expected.

When both primary (eg, shorter time to recovery) and secondary (eg, less progression to invasive mechanical ventilation or death) outcomes are met (ie, establishing a concordance in clinically relevant patient-centered outcomes), the results of a trial are swiftly translatable to clinical care. However, if the primary and secondary outcomes are discordant, as in the O'Halloran et al trial, then a different interpretation is needed. We propose the following interpretations of discordant (negative) primary and (positive) secondary outcomes.

No primary outcome will ever be perfectly designed during the planning of an RCT. That being said, a primary outcome that turns out to be negative should not be simply discarded; instead, it should be thoroughly investigated as to why it did not work. Explanation 1 is that it was not part of the expected biological-clinical rationale, explanation 2 is that the pretrial predicted primary outcome event rate was too low to detect any benefit, and explanation 3 is that a sufficient number of participants could not be recruited. It is critical to identify which of these 3 explanations led to the trial’s findings because the process of designing the next RCT testing these interventions should be informed by the final explanation. If the first explanation is correct, a different primary outcome that better matches the disease’s biological-clinical rationale will need to be incorporated. If the second explanation was correct, recalculating the outcome event rate based on more recent evidence will be necessary. If the third explanation was correct, an investigation of the barriers to participant recruitment will be needed to maximize feasibility.

While awaiting the results of the next RCT investigating the use of these immunomodulating agents for COVID-19, how should the current results be applied clinically?

These 2 drugs—abatacept and infliximab—that were associated with mortality reduction compared with placebo cannot be seen as equivalent to immunomodulators currently recommended by guidelines,9,10 even though the level of scientific evidence for currently standard immunomodulators differs (eg, multiple positive, placebo-controlled, double-masked RCTs for baricitinib11; multiple positive open-label but negative placebo-controlled RCTs for tocilizumab12; and 1 positive open-label RCT for dexamethasone13). Unless these drugs comprising the current immunomodulation of standard care are unavailable, abatacept and infliximab should not be routinely used for patients hospitalized for COVID-19. If a shortage of standard medications were to occur (as was the case during early 2022), abatacept and infliximab could be considered, especially given the known safety profile of these agents, even though their safety for use in COVID-19 remains limited. In the meantime, a replication of the trial findings by O'Halloran et al (ie, mortality reduction) is essential for abatacept or infliximab to have the potential to become part of standard care.

The study by O'Halloran et al is a well-executed, well-resourced trial that enhances the evidence around COVID-19 therapeutics and highlights challenges to the current scientific process, including strengths and flaws of trial design and interpretations and misinterpretations of scientific findings, all potentially associated with direct consequences for patients and society. The challenges of translating trial outcomes to clinical practice during pandemics require novel approaches and fresh solutions to improve the care of patients hospitalized for COVID-19, decrease treatment inequities, and ultimately optimize public health worldwide.

抱歉!评论已关闭.

×
腾讯微博